Count New Mexico child credit children under IRC 152(c); do not halve the 65+ medical exemption for separate filers

changelog.d/nm-ctc-152c-medical-exemption.fixed.md

@@ -0,0 +1 @@
+Count New Mexico child income tax credit children under IRC 152(c), so a child with an ITIN qualifies, and stop halving New Mexico's 65+ medical care expense exemption for married filing separately.

policyengine_us/tests/core/test_nm_ctc_children_medical_exemption_invariants.py

@@ -0,0 +1,377 @@
+"""New Mexico child income tax credit children and the 65+ medical exemption.
+
+NMSA 7-2-18.34(J)(2) defines the child income tax credit's "qualifying child"
+by IRC 152(c), which has no taxpayer identification number requirement, so
+`nm_ctc_qualifying_children` counts 152(c) children whatever their Social
+Security number status. NMSA 7-2-5.9 gives $3,000 to a filer or spouse 65 or
+older with at least $28,000 of medical care expenses (the PIT-ADJ reading of
+the threshold), with no halving for married filing separately.
+
+For households drawn by Hypothesis and a seeded population, in 2023-2026:
+
+1. Reference count: `nm_ctc_qualifying_children` equals an independent count
+   of the 152(c) tests the model records: a dependent who is the filer's
+   child (not a parent or grandparent) and is either permanently and totally
+   disabled, or younger than the older filer and under 19 or a full-time
+   student under 24. Outside New Mexico it is 0.
+2. Superset: where every dependent is younger than the older filer, it is
+   never below the federal `eitc_child_count`, which adds the Social Security
+   number requirement. (The federal helper does not apply the relative-age
+   test of IRC 152(c)(3)(A), so the two counts are not ordered otherwise.)
+3. Identification invariance: giving every dependent a Social Security
+   number changes neither the count nor `nm_ctc`.
+4. Label invariance: exchanging the head and spouse labels changes neither
+   the count, `nm_ctc` nor `nm_medical_expense_exemption`.
+5. Separate filing: a single adult filing separately gets half of `nm_ctc`
+   (7-2-18.34(G)) and the whole `nm_medical_expense_exemption`.
+6. Reference exemption: `nm_medical_expense_exemption` is $3,000 exactly when
+   a filer (not a dependent) is 65 or older and medical expenses are at least
+   $28,000, and 0 otherwise.
+"""
+
+import numpy as np
+import pytest
+from hypothesis import HealthCheck, given, settings
+from hypothesis import strategies as st
+
+from policyengine_us import Simulation
+
+YEARS = [2023, 2024, 2025, 2026]
+STATES = ["NM", "NM", "NM", "TX"]
+SSN_CARD_TYPES = ["CITIZEN", "NON_CITIZEN_VALID_EAD", "OTHER_NON_CITIZEN", "NONE"]
+RELATIONS = ["child", "child", "child", "parent", "grandparent"]
+EXEMPTION = 3_000
+MIN_EXPENSES = 28_000
+TOLERANCE = 0.01
+# Each drawn unit is computed four times: as drawn; with every dependent
+# given a Social Security number; with the head and spouse labels exchanged;
+# and filing separately (a single adult only).
+DRAWN, CITIZEN, SWAPPED, SEPARATE = range(4)
+
+
+def money(high):
+    return st.one_of(st.just(0), st.integers(1, high))
+
+
+@st.composite
+def adults(draw):
+    return {
+        "age": draw(st.integers(18, 90)),
+        "employment_income": draw(money(120_000)),
+        "claimed_as_dependent_on_another_return": draw(st.booleans()),
+    }
+
+
+@st.composite
+def dependents(draw):
+    relation = draw(st.sampled_from(RELATIONS))
+    if relation == "child":
+        age = draw(st.integers(0, 30))
+    else:
+        age = draw(st.integers(40, 95))
+    return {
+        "relation": relation,
+        "age": age,
+        "is_full_time_student": draw(st.booleans()) if 18 <= age <= 25 else False,
+        "is_permanently_and_totally_disabled": draw(
+            st.booleans() if relation == "child" else st.just(False)
+        ),
+        "ssn_card_type": draw(st.sampled_from(SSN_CARD_TYPES)),
+    }
+
+
+@st.composite
+def units(draw):
+    return {
+        "state": draw(st.sampled_from(STATES)),
+        "adults": draw(st.lists(adults(), min_size=1, max_size=2)),
+        "dependents": draw(st.lists(dependents(), max_size=3)),
+        "medical_expenses": draw(
+            st.one_of(
+                st.sampled_from([27_999, 28_000, 28_001]),
+                money(60_000),
+            )
+        ),
+    }
+
+
+def _seeded_units(n, seed=20261009):
+    rng = np.random.default_rng(seed)
+    out = []
+    for i in range(n):
+        drawn_adults = []
+        for _ in range(int(rng.integers(1, 3))):
+            drawn_adults.append(
+                {
+                    "age": int(rng.integers(18, 91)),
+                    "employment_income": float(round(rng.uniform(0, 120_000))),
+                    "claimed_as_dependent_on_another_return": bool(rng.random() < 0.2),
+                }
+            )
+        drawn_dependents = []
+        for _ in range(int(rng.integers(0, 4))):
+            relation = RELATIONS[int(rng.integers(len(RELATIONS)))]
+            age = int(
+                rng.integers(0, 31) if relation == "child" else rng.integers(40, 96)
+            )
+            drawn_dependents.append(
+                {
+                    "relation": relation,
+                    "age": age,
+                    "is_full_time_student": bool(
+                        18 <= age <= 25 and rng.random() < 0.5
+                    ),
+                    "is_permanently_and_totally_disabled": bool(
+                        relation == "child" and rng.random() < 0.1
+                    ),
+                    "ssn_card_type": SSN_CARD_TYPES[
+                        int(rng.integers(len(SSN_CARD_TYPES)))
+                    ],
+                }
+            )
+        out.append(
+            {
+                "state": STATES[i % len(STATES)],
+                "adults": drawn_adults,
+                "dependents": drawn_dependents,
+                "medical_expenses": float(
+                    [27_999, 28_000, 28_001][i % 3]
+                    if i % 4 == 0
+                    else round(rng.uniform(0, 60_000))
+                ),
+            }
+        )
+    return out
+
+
+def _crafted_units():
+    """Edge cases the seeded draw rarely reaches."""
+
+    def adult(age, income=20_000.0):
+        return {
+            "age": age,
+            "employment_income": income,
+            "claimed_as_dependent_on_another_return": False,
+        }
+
+    def dependent(age, student=False, disabled=False, ssn="OTHER_NON_CITIZEN"):
+        return {
+            "relation": "child",
+            "age": age,
+            "is_full_time_student": student,
+            "is_permanently_and_totally_disabled": disabled,
+            "ssn_card_type": ssn,
+        }
+
+    def unit(adults, dependents, expenses=0.0):
+        return {
+            "state": "NM",
+            "adults": adults,
+            "dependents": dependents,
+            "medical_expenses": expenses,
+        }
+
+    elderly_parent = {**dependent(70, ssn="CITIZEN"), "relation": "parent"}
+    return [
+        # A student sibling older than, or as old as, the filer.
+        unit([adult(20)], [dependent(22, student=True)]),
+        unit([adult(20)], [dependent(20, student=True)]),
+        # A disabled sibling older than the filer.
+        unit([adult(20)], [dependent(30, disabled=True)]),
+        # Younger than the older spouse only.
+        unit([adult(25), adult(20)], [dependent(22, student=True)]),
+        # The only 65+ person is a dependent parent.
+        unit([adult(40)], [elderly_parent], expenses=30_000.0),
+        # A 65+ filer with exactly $28,000 of expenses.
+        unit([adult(70)], [], expenses=28_000.0),
+    ]
+
+
+def _situation(drawn_units, year):
+    people, tax_units, households = {}, {}, {}
+    for i, unit in enumerate(drawn_units):
+        for copy in (DRAWN, CITIZEN, SWAPPED, SEPARATE):
+            members = []
+            adult_list = list(unit["adults"])
+            if copy == SWAPPED:
+                adult_list = adult_list[::-1]
+            for j, adult in enumerate(adult_list):
+                name = f"adult_{i}_{copy}_{j}"
+                people[name] = {
+                    "age": adult["age"],
+                    "employment_income": adult["employment_income"],
+                    "claimed_as_dependent_on_another_return": adult[
+                        "claimed_as_dependent_on_another_return"
+                    ],
+                    "is_tax_unit_head": j == 0,
+                    "is_tax_unit_spouse": j == 1,
+                    "is_tax_unit_dependent": False,
+                    "medicare_enrolled": False,
+                }
+                members.append(name)
+            people[members[0]]["other_medical_expenses"] = unit["medical_expenses"]
+            for j, dependent in enumerate(unit["dependents"]):
+                name = f"dependent_{i}_{copy}_{j}"
+                people[name] = {
+                    "age": dependent["age"],
+                    "is_full_time_student": dependent["is_full_time_student"],
+                    "is_permanently_and_totally_disabled": dependent[
+                        "is_permanently_and_totally_disabled"
+                    ],
+                    "is_parent_of_filer_or_spouse": dependent["relation"] == "parent",
+                    "is_grandparent_of_filer_or_spouse": dependent["relation"]
+                    == "grandparent",
+                    "ssn_card_type": (
+                        "CITIZEN" if copy == CITIZEN else dependent["ssn_card_type"]
+                    ),
+                    "is_tax_unit_head": False,
+                    "is_tax_unit_spouse": False,
+                    "is_tax_unit_dependent": True,
+                    "medicare_enrolled": False,
+                }
+                members.append(name)
+            tax_unit = {"members": members}
+            if len(unit["adults"]) == 2:
+                tax_unit["filing_status"] = {year: "JOINT"}
+            elif copy == SEPARATE:
+                tax_unit["filing_status"] = {year: "SEPARATE"}
+            tax_units[f"tax_unit_{i}_{copy}"] = tax_unit
+            households[f"household_{i}_{copy}"] = {
+                "members": members,
+                "state_code": {year: unit["state"]},
+            }
+    people = {
+        name: {k: {year: v} for k, v in values.items()}
+        for name, values in people.items()
+    }
+    return {"people": people, "tax_units": tax_units, "households": households}
+
+
+def _reference_children(unit):
+    if unit["state"] != "NM":
+        return 0
+    oldest_filer = max(a["age"] for a in unit["adults"])
+    return sum(
+        d["relation"] == "child"
+        and (
+            d["is_permanently_and_totally_disabled"]
+            or (
+                d["age"] < oldest_filer
+                and (d["age"] < 19 or (d["is_full_time_student"] and d["age"] < 24))
+            )
+        )
+        for d in unit["dependents"]
+    )
+
+
+def _reference_exemption(unit):
+    if unit["state"] != "NM":
+        return 0
+    filer_65 = any(a["age"] >= 65 for a in unit["adults"])
+    return EXEMPTION * (filer_65 and unit["medical_expenses"] >= MIN_EXPENSES)
+
+
+def _describe(unit):
+    return (
+        f"{unit['state']} adults={[(a['age'], a['claimed_as_dependent_on_another_return']) for a in unit['adults']]} "
+        f"dependents={[(d['relation'], d['age'], d['ssn_card_type']) for d in unit['dependents']]} "
+        f"expenses={unit['medical_expenses']}"
+    )
+
+
+def check_invariants(drawn_units, year):
+    sim = Simulation(situation=_situation(drawn_units, year))
+
+    def calc(name):
+        return np.asarray(sim.calculate(name, year), dtype=float).reshape(-1, 4)
+
+    children = calc("nm_ctc_qualifying_children")
+    eitc_children = calc("eitc_child_count")
+    credit = calc("nm_ctc")
+    exemption = calc("nm_medical_expense_exemption")
+    for i, unit in enumerate(drawn_units):
+        where = f"{year} unit {i}: {_describe(unit)}"
+        single = len(unit["adults"]) == 1
+        # 1. Reference count.
+        assert children[i, DRAWN] == _reference_children(unit), where
+        # 2. Superset of the EITC count where every dependent is younger than
+        # the older filer (the New Mexico count is 0 elsewhere).
+        oldest_filer = max(a["age"] for a in unit["adults"])
+        if unit["state"] == "NM" and all(
+            d["age"] < oldest_filer for d in unit["dependents"]
+        ):
+            assert children[i, DRAWN] >= eitc_children[i, DRAWN], where
+        # 3. Identification invariance.
+        assert children[i, CITIZEN] == children[i, DRAWN], where
+        assert abs(credit[i, CITIZEN] - credit[i, DRAWN]) <= TOLERANCE, where
+        # 4. Label invariance.
+        assert children[i, SWAPPED] == children[i, DRAWN], where
+        assert abs(credit[i, SWAPPED] - credit[i, DRAWN]) <= TOLERANCE, where
+        assert abs(exemption[i, SWAPPED] - exemption[i, DRAWN]) <= TOLERANCE, where
+        # 5. Separate filing.
+        if single:
+            assert abs(credit[i, SEPARATE] - credit[i, DRAWN] / 2) <= TOLERANCE, where
+            assert abs(exemption[i, SEPARATE] - exemption[i, DRAWN]) <= TOLERANCE, where
+        # 6. Reference exemption.
+        assert abs(exemption[i, DRAWN] - _reference_exemption(unit)) <= TOLERANCE, where
+    return credit[:, DRAWN], exemption[:, DRAWN]
+
+
+def _assert_coverage(drawn_units, credit, exemption):
+    """The seeded population reaches the cases the invariants are about."""
+    single = np.array([len(u["adults"]) == 1 for u in drawn_units])
+    child_without_ssn = np.array(
+        [
+            any(
+                d["relation"] == "child"
+                and d["age"] < 19
+                and d["ssn_card_type"] in ("OTHER_NON_CITIZEN", "NONE")
+                for d in u["dependents"]
+            )
+            for u in drawn_units
+        ]
+    )
+    elderly_dependent_only = np.array(
+        [
+            u["state"] == "NM"
+            and all(a["age"] < 65 for a in u["adults"])
+            and any(d["age"] >= 65 for d in u["dependents"])
+            and u["medical_expenses"] >= MIN_EXPENSES
+            for u in drawn_units
+        ]
+    )
+    not_younger_than_filers = np.array(
+        [
+            any(
+                d["relation"] == "child"
+                and not d["is_permanently_and_totally_disabled"]
+                and d["age"] >= max(a["age"] for a in u["adults"])
+                and (d["age"] < 19 or (d["is_full_time_student"] and d["age"] < 24))
+                for d in u["dependents"]
+            )
+            for u in drawn_units
+        ]
+    )
+    assert (child_without_ssn & (credit > 0)).any()
+    assert not_younger_than_filers.any()
+    assert (single & (credit > 0)).any()
+    assert (single & (exemption > 0)).any()
+    assert elderly_dependent_only.any()
+
+
+@settings(
+    max_examples=5,
+    deadline=None,
+    derandomize=True,
+    suppress_health_check=[HealthCheck.too_slow, HealthCheck.data_too_large],
+)
+@given(st.sampled_from(YEARS), st.lists(units(), min_size=8, max_size=20))
+def test_nm_ctc_children_and_medical_exemption_hypothesis(year, drawn_units):
+    check_invariants(drawn_units, year)
+
+
+@pytest.mark.parametrize("year", YEARS)
+def test_nm_ctc_children_and_medical_exemption_seeded_population(year):
+    drawn_units = _seeded_units(40) + _crafted_units()
+    credit, exemption = check_invariants(drawn_units, year)
+    _assert_coverage(drawn_units, credit, exemption)